Date: Wed, 2 Dec 2009 08:12:25 -0800
From: melvyn wright 
To: Jun-Hui Zhao 
Cc: Peter Teuben , dick plambeck ,
     Statia Luszcz 
Subject: Re: smauvplt.for

Hi Jun-Hui,

thanks for  SMA recipes and documentation pointers for low SNR
bandpass calibration.  we'll give these a try on some Carma data.

Mel.

On Wed, Dec 2, 2009 at 7:39 AM, Jun-Hui Zhao  wrote:
> Hi Mel,
> I interleave my thoughts in your comments below.
>  Jun-Hui
>
>
> On Tue, 1 Dec 2009, melvyn wright wrote:
>
>> Hi Jun-Hui,
>>
>> OK. thanks. In principle, setting MAXWIN=98 should
>> work, but I would not be suprised if some format and plotting
>> tasks need to be modified to handle that.
> I agree with you. So far, I just implemented the smalod for
> load the 4GHz x 2sb data into Miriad. I will look into calibration
> and other programs in Miriad for handling SMA DB data.
>
>>
>> 2) SMAMFCAL -- A separate question.
>>
>> What is the procedure that SMA uses for
>> low SNR narrow bandpass determination.
>>
>> Is there an SMA recipe for deriving bandpass and gains
>> for narrow spectral lines.
>> Do you have a favored combination of weight= xxx
>> and options=interpolate ?
>
> I don't have a particular procedure for a low S/N source.
> But the default setup in SMAMFCAL with options=msmooth, i.e.
> weight=2 and smooth=3,1,0.9, gives a good solution
> for most of the cases in handling SMA low S/N data.
>
> weight=2 is always good for resolved planets and
> high-angular-resolution data of QSOs with poor coherent long-baselines.
>
> If one chooses the moving smooth (options=msmooth) to improve S/N, the
> first parameter k in smooth (2k+1 is the number of channels involved in
> the moving smooth window) is sensitive, which depends upon the actual
> spectral resolution and S/N of each channel, i.e. for a very high spectral
> resolution and poor S/N, one has to try a large k.

>
> If one chooses orthogonal polynomial LSQ fitting to improve S/N,
> I found polyfit=3 (cubic), default selection, appears to be good.
> I haven't tested the high orders yet.
>
> The brief descriptions in choosing these parameters are given in the
> inline document along with SMAMFCAL. In the SMA data reduction guide,
> several cases in handling bandpass based on the procedures used by students,
> postDr. and scientists in the past are given
> http://www.cfa.harvard.edu/sma/miriad/manuals/SMAuguide/smauserhtml/node84.html
>
> I just went back to read them and found that these cases are helpful but
> didn't explain clearly why choose the particular setups. I would like to
> improve the recipes based on experiements with a low S/N high angular
> resolution observation at a high frequency.
>
> Jun-Hui
>